Study On Key Technologies For Trust-based Web Service Composition

With the rapid development of network, communication and software technology, the Internet of Things (Abbr. IoT), as another breakthrough in information technology, has become key technology in modern information science. IoT is referred to as a network that connects every object in the real-world through different kinds of wireless sensors according to the agreed protocols, and enables the exchanges of information and communication to achieve intelligent identification, registration, tracking and management. The development of sensing devices also leads to the software development in the IoT’s application layer. In order to shield the software underlying differences and improve the reusability of software in different environments, modular software developments have become the hot points in application layer software. In this structure, the software modules in different organizations and platforms are published as services to coordinate and interact with other software applications, which can effectively reduce the development difficulties caused by the complexity of software functions and structures. However, the dynamic and diversity characteristics of open networks and the rapid increasing of service resources cause the trustworthy service selection and composition a critical problem in service-oriented computing environments.Therefore, this dissertation mainly studies the key technologies for Web services composition based on trust evaluation in different scenarios, and gets the following results:1. We first work on the time-constrained trust-based service composition issue in multi-domain environments. In the multi-domain environments, the great expense of interdomain validations, the dynamic execution time and the subjective requirement for global execution time remain serious obstacles to the trust-based service composition. To solve these issues, we proposes a novel service-composition approach, modeling the trust-based service composition as the multi-domain scheduling and assignment problem using the minimum service resources within a certain time constraint. In our approach, each service is designed as an exclusive resource during its execution. By analyzing the required interdomain communications, the number of available services and the aggregated trust value in each domain, the optimal domain can be obtained at each step of service selection. Meanwhile, loop parallelization and predecessor reselection are adopted when the task cannot be completed on schedule. Finally, to raise higher trust value of the composite result, the redundant service resources of the initial composition are further optimized. The experimental results illustrate that the proposed trust-based service-composition approach outperforms traditional ones in both effectiveness and scalability under various time constraints.2. We also work on the trust-based service composition issue in service-oriented mobile social networks. In service-oriented mobile social networks, it is difficult to compose trustworthy services because of the opaque characteristics of service components, the subjective interpretation of trust requirements, and the opportunistic connectivity of social participants. To address these issues, we propose a trust-based decentralized service-composition approach in service-oriented mobile social networks. Our approach decentralizedly evaluates the trustworthiness of target services at both component and composite levels based on a lattice-based trust model to prevent data from being transmitted to untrustworthy counterparts. Moreover, by analyzing the dependency relationship, service components are selected satisfied both global and local constraints specified by service consumers and vendors respectively. Finally, based on Monte Carlo method and its optimization, we further develop trustworthy social path selection and fault-tolerant approaches to deliver service evaluation and invocation information. Experimental results demonstrate its practical suitability of service composition for service-oriented mobile social networks, which leads to low cost for service evaluation, as well as high success rate and aggregated trust value for path selection.3. And we work on the fairness-guaranteed reputation propagation issue for Web service composition. In Web service composition scenario, the evaluation of component services is difficult because of the opaque characteristic of composite services, the complex invocation structures, and the subjective reputation rating of service consumers. To address these issues, this dissertation proposes a reputation propagation algorithm for Web service composition, in which the subjective ratings can be fairly propagated to each component service. The algorithm first models service composition as the Beta-mixture to learn the reputation and responsibility of each component through EM algorithm. Then, based on the characteristics of Shapley values in cooperative gaming theory, the algorithm computes the contribution of each component to its composition, enabling no component would obtain extra rewards or punishments. Finally, the theoretical analysis and the experimental results demonstrate the fairness of the algorithm to hieratically propagate consumer’s rating to each component service.